If you were to look out your at this moment, you might see a landscape that includes a number of trees, a few buildings, pedestrians on the road and a dustbin. Turning your gaze the other direction, you may see a lake, a bus stop, a dog, and an arriving Bus. Even though you are seeing objects, what you see in each view is more than just a group of objects—it is a composite item we label a “scene.”.
Object recognition and face recognition algorithms have reached near human performance in good lighting.
Till a while ago, computer vision was largely limited to recognizing individual objects. Current advances in artificial intelligence have created software capable of recognizing and describing the content of photographs and videos with far greater accuracy than ever before, sometimes even mimicking human levels of understanding.
This new class of software called “Scene Recognition” can teach itself to identify entire scenes: a group of young men playing football, for example, or a group of pedestrians waiting to cross the road. The software then writes a caption in English describing the picture. Compared with human observations the computer-written descriptions are surprisingly accurate.
With scene recognition, you’re trying to tell the story behind the image. A natural scene will be very complex, and you want to pick out the most important objects in the image. Aido uses libraries from a leading Vision Vendor to do scene recognition.